J. Phys. Chem. 1983, 87, 4616-4618
4616
100.-
for which the upper state curvature is negative, corresponding to the imaginary frequency wlm' = 1560 i cm-'. The upper state potential energy surface has a local maximum or saddle point a t the geometry for which dissociation forming CO(lZ') S(lD) or CS('Z+) O(lD) are equally likely. The negative curvature derived from the observed shift implies that the Franck-Condon region of the upper state lies at an energy above the inflection between this local maximum and the positive curvature where the dissociation products are still further separated. The solid curve associated with the azzz(v)data in Figure 1 is calculated for the data at 403.6 K. It has the same shape as the aW(u) curve. Its spectral moment is shifted in accordance with the composite population of states excited in the CS stretching mode. Its band strength is normalized to agree with the composite bending mode population, as given by eq 3. The agreement between the arzz(v)data points and the solid curve constructed in this way is rough, mainly because populations on which they are based are small. The small, but systematic component of a,,,(v) strength a t the high frequency end of the spectrum could result from a small errror in the approximation that aW(u) and aolo(u)are identical except for their shift and relative strength. In this case, the moment analysis is nevertheless correct, and the solid curve is the best representation of arz2(v)available short of a more detailed account of the shapes of each component. In any case, it seems clear that the basis for the overall shape of the spectrum cannot be a long progression in either the upper state bending or CS stretching modes, in which case it must be a long upper state progression in the CO stretching mode.
+
-
I
E
0
Y
0 0
>N50.-
a
7 .06
.04
.02 fi
Figure 2. Plot of residual spectral shift vs. fractional number of C-S stretching quanta as defined in the text. The slope of this plot is the spectral shift A V , due ~ to excitation of the C-S stretch, and is equal to 1840 cm-'.
tributions to the shift are additive, Avzoois related to Avow by Avz00(?? = Avobscj(??CO' + 1)pijk(n ij k
CG + l)AJ'ojOpijk(T) = Avlo~CiO'+ 1)pijk(T) (9) ijk
ijk
where AviW = iAvlo0.9 The slope of a plot of Av,,(r) as a function of CiG l ) P i j k ( nas, shown in Figure 2, gives the shift of the CS stretching component AuIm = 1840 cm-'. For a vibrational mode which does not change the symmetry of the molecule, the shift is given by the sum rule
+
(9) S. M.Adler-Golden, Chem. Phys., 64, 421 (1982).
+
Acknowledgment. We acknowledge the support of the Naval Weapons Support Center, Crane, Indiana through Naval Air Systems Command AIR32R in carrying out the research. Registry No. Carbonyl sulfide, 463-58-1.
Phase Diagram of y-Oxygen Junilng Yen and Malcolm Nlcol' Department of Chemistry and Biochemistry. University of California, Los Angeles, California 90024 (Received: April 15, 1983)
The phase diagram of 7-0,has been determined by observing Raman spectra of the 0, stretching mode at temperatures from 120 to 300 K and pressures between 1and 6 GPa. The y phase was identified by the splitting of the Raman spectrum of the stretchingmode into two components which differs from the single-peaked spectra of the 8 and fluid phases. Extrapolations of the 7-0,boundaries determined by this method closely match the boundaries reported by Stevenson and Stewart at lower temperatures and pressures. The triple point between yo2,P-O2,and fluid O2 was located at 5.0 A 0.3 GPa and 283 f 4 K. The relatively narrow stability field of 7-0, is attributed to the instability of this relatively disordered structure with respect to structures in which the incompletely filled s* orbitals of adjacent oxygen molecules overlap to form stable chemical bonds.
The solid homonuclear diatomics H2, N2, O2 and the halogens are in many respects the simplest molecular crystals. Thus, in order to develop a detailed understanding of the forces that control the properties of mo0022-3654/83/2087-46 16$01.50/0
lecular crystals, it is of interest to know how the structures, dynamics, and relative stabilities of these solids vary with molecular properties (permanent electrical moments or more detailed aspects of molecular electronic structure), 0 1983 American Chemical Society
Phase Diagram of ?-Oxygen
intermolecular separation, and orientation. N2, 02,and F2 often are considered to form qualitatively similar crystals. Indeed, several polymorphs of each system are isomorphous with at least one polymorph of another system. However, these parallels are only poor first approximations. For example, our earlier studies of the 0, system showed1q2that yo2,which crystallizes from the liquid at 55 K under atmospheric p r e s s ~ r e s ,does ~ ? ~ not occur at room temperature. Other preliminary observations suggested that, at any temperature, 7-0,has a relatively narrow field of stability, less than 1GPa, whereas the isomorphous high-pressure phase of N2 appears to be stable over as many as 30 GPa at room temperature.6,6 This difference was one of the motivations for the effort to define the stability field of y-0, reported here. Oxygen differs from other homonuclear diatomics in another important respect; its electronic ground state is an open-shell triplet. In the a phase which is stable below 23 K at atmospheric pressure, the spin system has longrange antiferromagnetic order. Magnetization and neutron scattering measurements for polycrystalline 0-0, suggest that magnetic order persists in this phase, which is stable between 23 and 44 K at 1atm.7 However, the first-order transition near 44 K to the y phase, which is stable between 44 and 55 K, and the large differences between the y- and 0-0, structures prevent growth at low pressures of p-0, crystals of a size sufficient for the diffraction studies that could characterize the detailed magnetic order of the a phase. A secondary objective of this study was to determine the minimum pressure at which single crystals of 0-0, could be grown from the melt for neutron diffraction or magnetization studies. Experimental Section Oxygen samples were loaded in a diamond-anvil cell of Holzapfel's design8 by the method used for the earlier studies.l The cell was mounted in a variable-temperature cryostat similar to that described by Hirsch and Holzapfel.* Their design was modified, however, to compensate for thermal contraction of the liquid-nitrogen reservoir by expansion of metal bellows which connect the reservoir to the head of the cryostat, and to support the cell from below by a Fiberglas-epoxy composite rod. These modifications kept the sample at the focus of an optical system during thermal cycling without external adjustments. The temperature of the sample was measured by a Pt resistance thermometer located as close to the gasket as possible. Pressures were determined by comparing the R1 luminescence spectrum of a ruby chip mounted in the sample with the spectrum of another chip of the same ruby under vacuum at the same temperature. During pressure measurements, the output power of the Ar laser used to excite the ruby was maintained well below 50 mW; the power at the sample was lower than this by at least a factor of 2. At these powers, the ruby spectra did not shift with power, which suggests that the ruby and the sample were not being heated above the temperature of the cell. The factor 0.365 nm/GPa was used to convert the shift of the R1 line to pressure. (1) Nicol, M.; Hirsch, K. R.; Holzapfel, W. B. Chem. Phys. Lett. 1979, 68, 49-52. (2) d'Amour, H.; Holzapfel, W. B.; Nicol, M. J. Phys. Chem. 1981,85, 13C-1. (3) Stevenson, R.J. Chem. Phys. 1957,27,673-5. (4) Stewart, J. W. J. Phys. Chem. Solids 1959, 12, 122-9. (5) LeSar, R.; Ekberg, S. D.; Jones, L. H.; Mills, R. L.; Schwalbe, L. A.; Schiferl, D. Solid State Commun. 1979, 32, 131-4. (6) Cromer, D. T.; Mills, R. L.; Schiferl, D.; Schwalbe, L. A. Acta Crystallogr., Sect. B 1981, 37, 8-11. (7) deFoitis, G. Phys. Reu. 1981, 823, 4714-40. (8) Hirsch, K. R.; Holzapfel, W. B. Reu. Sci. Instrum. 1981,52, 52-5.
The Journal of Physical Chemistry, Vol. 87, No. 23, 1983 4617
I !\
i\ 1555
1560
1565
1570
1555
1560
1565
Y-0,
1570
WAVENUMBER/CM-'
Flgure 1. Raman spectra of the stretching mode of 0,obtained while a sample of p-0, at 172 K and 2.2 GPa was warmed through the 6-7 and y-fluid transitions.
Phase transitions were detected by the combination of visual observations of the sample while the cell was heated or cooled under constant applied load and of Raman spectroscopy excited with Ar 514.5-nm or Kr 649.1-nm excitation. In transmitted light, the P-to-y boundary is evident by the transformation from pink to virtually colorless crystals, but the transformation between the colorless y and fluid phases is more difficult to observe. Thus, Raman spectra were particularly helpful for determining the melting curve. The visual observations and the Raman studies with red excitation also circumvent problems of sample heating that must be considered with green excitation. In this study, identical transition conditions were observed for all three detection modes. The Raman spectra also appeared to be independent of the laser power (up to -0.8 W at the output mirror), even near the 8-7 boundary. Together with the fact that visual examination of the 0, samples before and after exposure to the laser excitation provided no evidence of recrystallization, these observations suggest that artifacts from laser heating were avoided. Results The pale colors of 7-0, and fluid 0, are very similar; thus, it is difficult to determine the melting line by direct observation. However, these phases can be clearly distinguished by Raman spectroscopy. The Raman spectra of fluid 0, and p-0, in the 0-0 stretching region (1550-1600 cm-l) consist of single bands which have different bandwidths1 and, at any pressure, slightly different spectrum, however, consists of two frequencies. The 7-0, bands that are separated from each other by between 2 cm-' near 1 GPa to 5 cm-' near 5 GPa. A t any pressure, both bands of the 7-0, spectrum are shifted slightly from
4618
The Journal of Physical Chemistry, Vol. 87, No. 23, 1983
+ looY
Yen and Nicol
I
I
4
I
P’d 0
2.0
4.0
6.0
PRESSURE/ GPa
Flgure 2. Partial phase diagram of 0, near the stability field of 7-02.
the fluid and p bands. Thus, as Figure 1 shows for a warming cycle initiated at 2.2 GPa and 172 K, the single band of the Raman spectrum of p-0, below 175 K is replaced discontinuously on warming by two bands of the spectrum of 7-0, at 2.7 GPa and 185 K; and this is replaced by the single broad band of the fluid spectrum at 3.2 GPa and 236 K. Similar transformations with little hystersis are observed along cooling or approximately isothermal compression paths. From these and similar observations at other pressures, the p-y and 7-fluid phase boundaries depicted in Figure 2 by open squares and open circles, respectively, were determined. When these boundaries are extrapolated to lower pressures, they overlap reasonably well with data reported by Stevenson3 and S t e ~ a r t . ~ When oxygen samples were warmed through a very small region, about 5.0 f 0.3 GPa and 283 f 4 K, a three-peaked spectrum emerged from the single-band spectrum of 6-0, instead of the double-banded spectrum of yo2. On further heating, these three peaks were immediately replaced by the single-banded spectrum of the fluid, as shown in Figure 3. A t very slightly higher pressures, neither the double-peakedspectrum of 7-0,nor the triple-peaked spectrum could be detected. At best, coexistence of bands associated with p and the fluid was observed. Thus, within the precision of these measurements, f0.3 GPa and f4 K, the triple point between the 6, y, and fluid phases of O2 is at 5.90 GPa and 283 K. This pressure is unfortunately high relative to the design limitations of high-pressure cells for neutron diffraction. The limited stability of 7-0,restricta the precision with which the dependence of the factor group splitting of the internal mode on pressure or temperature can be measured and made it impossible to separate the temperature and pressure dependences. However, the two bands appear to move apart by about 0.1 cm-’ GPa-’. If this is primarily a pressure dependence, it is essentially identical with the pressure dependence of the factor-group splitting of the internal mode of the isomorphous phase of N2 increases with p r e s ~ u r e .Furthermore, ~ at any pressure, the factor-group splittings of both isomorphs are essentially equal. To this degree, the intermolecular potentials for the two systems appear to be similar at large intermolecular separations. Figure 4 shows a plot of the splitting of the Raman spectrum of the stretching mode of 7-0,vs. pressure. As these systems are compressed, however, O2molecules, with their open-shell ground states, can lower their energies
7
1560
1565
1570
1575
1580
WAVENUMBER/CM-’
Figure 3. Raman spectra of the stretching mode of 0, at pressures near 5.0 GPa obtained while a sample of the p phase at 270 K was warmed and melted. The three-peaked spectrum at 5.0 GPa and 283 K is interpreted in terms of coexistence of 8- and 7-0,and fluid 02. Pressures for the 270 and 278 K spectra were not measured during this warming run, but measurements at lower temperatures imply that the corresponding pressures were within 0.2 GPa of 5.0GPa.
I
I
1.01 La 0 2.0
1
4.0
6.0
PRESSURE/GPa
F M r e 4. Plot of the splltting of the Raman spectrum of the stretching mode of y-O2 vs. pressure.
in a manner that is not available to closed-shell molecules such as N2and, presumably, F,. That is, by transforming to structures in which the incompletely filled orbitals on adjacent molecules overlap, chemical bonds can be formed between O2 molecules. a-,p-, and t-02and orthorhombic O2are such structures. Their stabilities in the O2 system appear to be a consequence of the specific open-shell electronic structure of the O2 molecule. Acknowledgment. It is a pleasure to acknowledge valuable assistance provided by Ms. Andrea Schwake. This work was supported by National Science Foundation Grant DMR 80-25620. Registry No. Oxygen, 1182-44-1.